2409
Anal. Chem. 1993, 65, 2489-2492
Migration Behavior of Phthalate Esters in Micellar Electrokinetic
Chromatography with or without Added Methanol
Sahori Takeda,' Shin-ichi Wakida, Masataka Yamane, Akinori Kawahara, and Kunishige Higashi
Department of Material Chemistry, Government Industrial Research Institute, Osaka, Midorigaoka 1-8-31,
Ikeda, Osaka 563, Japan
Selected phthalate esters including priority pollutants were analyzed by micellar electrokinetic
chromatography (MEKC). Addition of methanol
(20%, v/v) to aqueous migration buffer solution
containing 0.05 M sodium dodecyl sulfate (SDS)
improved the resolution of their separation. To
investigatethe migration behavior of the phthalate
esters, we calculated distribution coefficients in
micellar solubilizationof the phthalate esters from
the analytical data of MEKC. A linear relationship between the logarithms of the measured
distribution coefficients and those of octanolwater partition coefficients reported in the literature was observed with both SDS solution and
methanolmixed solution. Using this relationship,
the migration times of phthalate esters were
estimated from their octanol-water partition coefficients. Enthalpy and entropy changes resulting from their micellar solubilization were also
evaluated in order to investigate the distribution
mechanism of the phthalate esters. The enthalpy
changes decreased with an increase in the alkyl
chainlengthof thephthalateesters with both SDS
solution and methanol mixed solution. On the
other hand,the trends of theentropy changes with
alkyl chain length were different for the two
solutions. These results suggest that the solutesolvent interaction is different for the two solutions.
INTRODUCTION
Micellar electrokinetic chromatography (MEKC) is a new
analytical method which can provide high resolution.1.2 This
method is performed with the same apparatus as capillary
electrophoresis although its separation principle is based on
chromatography, that is, the difference in the distribution
between solvent and micelles of ionic surfactants. In MEKC,
it is possible to estimate distribution coefficients and thermodynamic parameters in micellar solubilization.2
For target analytes with MEKC, we chose phthalate esters.
Phthalate esters are widely used as plasticizers in industry,
so they are common contaminants in aquatic environment^.^
According to the US. Environmental Protection Agency
(EPA) method 606,4 dimethyl phthalate (DMP), diethyl
phthalate (DEP),di-n-butyl phthalate (DNBP), benzyl butyl
phthalate (BBP), bis(2-ethylhexyl) phthalate (DEHP), and
di-n-octyl phthalate (DNOP) are listed as the priority
pollutants among phthalate esters. EPA method 606 specifies
analysis by gas chromatography with an electron capture
detector; however, appropriate pretreatment is necessary. In
some cases, phthalate esters are analyzed by high-performance
liquid chromatography (HPLC) but the resolution is poor.6
The separation of some phthalate esters by MEKC with
SDS solution has been reported.6 Six selected phthalate esters
were separated, and it was pointed out that the migration
order agreed with that of the logarithms of octanol-water
partition coefficients. However, the resolution of phthalate
esters was poor and the octanol-water partition coefficients
were simply used to qualitatively explain the migration order
without quantitative discussion. In this paper, we present
the quantitative relationship between distribution coefficients
in micellar solubilization determined by MEKC with SDS
solution or methanol mixed solution and octanol-water
partition coefficients reported in the literature.' Also a
method of estimating the migration time of phthalate esters
from this relationship is described. The thermodynamic
parameters in micellar solubilization of phthalate esters are
determined by MEKC with both SDS solution and methanol
mixed solution. The trends of thermodynamic parameters
in SDS solution or methanol mixed solutions are discussed.
EXPERIMENTAL SECTION
Apparatus. MEKC was performed with a Model 270A
analyticalcapillary electrophoresis system (Applied Biosystems
Inc.). A fused-silica capillary tube (720 mm X 50 pm i.d.) was
used as a separation tube. Both ends of the tube were dipped
into carrier solution with platinum electrodes and dc voltage
(5-30 kV) was applied between them. The capillary was
thermostated by air coolant. Migrating solute bands were
detected at 500 mm from the positive end by on-column
measurementof UVabsorption (210nm). A ChromatopacC-R6A
(Shimadzu) was used for data processing.
Reagents. Selected phthalate esters, DMP, DEP, DNBP,
BBP, DEHP, DNOP, and diisobutyl phthalate (DIBP) were
obtained from Wako Pure Chemicals. Di-n-propyl phthalate
(DNPP),diisopropyl phthalate (DIPP),and di-n-amylphthalate
(DNAP) were obtained from Tokyo Kasei Kogyo. Sodium
dodecyl sulfate from Nacalai Tesque was used as a micelle-forming
anionic surfactant. An SDS solution was prepared by dissolving
SDS in a mixture of 0.02 M sodium dihydrogen phosphate solution
and 0.02 M sodium tetraborate solution adjusted to pH 9.0. All
reagents and solvents were of analyticalgrade and used without
further purification.
Procedure. Stock solutions of the phthalate esters were
prepared to be 10 g/L in methanol. A standard solution of a
mixture of 10 phthalate esters was made by diluting the stock
solutionswith SDSsolution. The concentration of each phthalate
ester was 50 mg/L, except for DEHP and DNOP, for which the
concentration was 25 mg/L. Samples were injected by vacuum
injection (5 in. of Hg, 0.2 8 ) throughout all experiments. The
injection volume was about 1.5 nL.
~~~
(1) Terabe, S.;Otsuka, K.; Ichikawa, K.; Tsuchiya, A.; Ando, T. Anal.
Chem. 1984,56, 111.
(2) Terabe, S.;Otsuka, K.; Ando, T. Anal. Chem. 1986,57, 834.
(3) Wolfe, N. L.; Burns, L. A.; Steen, W. C. Chemosphere 1980,9,393.
(4) Keith, L. H.; Telliard, W. A. Enuiron. Sci. Technol. 1979,13,416.
0003-2700/93/0365-2489$04.00/0
(5) Mori, S. J. Chromutogr. 1976, 129, 53.
(6) Ong, C. P.; Lee, H. K.; Li, S. F. Y. J. Chromatog. 1991,542,473.
(7) Leyder, F.; Boulanger, P. Bull. Enuiron. Contam. Toxicol. 1983,
30, 152.
0 1993 Amerlcan Chemlcal Society
2490
ANALYTICAL CHEMISTRY, VOL. 65, NO. 18, SEPTEMBER 15, 1993
mechanism of separation is almost maintained, as evidenced
by the experimental data and from the following discussion.
Relationshipof Distribution Coefficients to OctanolWater Partition Coefficients. To investigate the distribution of phthalate esters with both SDS solution and
methanol mixed solution, the distribution coefficients in
micellar solubilization were calculated from analytical data
as follows.2 The capacity factor &' can be calculated from
migration time of water to, solute t ~and
, micelle t,, by
BBP
I
k'
I
0
tmc
\:
20
j10
30
I
160
50
40
,
Time/min
1
0
:
I
I
1
1
1
2
I
5
10
i?
2
I
t
I
1
0 / p
50 100
Figure 1. Chromatogramsof phthalate esters in MEKC wlth methanol
mixed solution. Conditions: micellar solution, 0.05 M SDS in 0.02 M
borate-phosphate buffer solution(pH9.O)-methanol(8020, v/v); applied
voltage, 20 kV; temperature, 38.9 OC; detection wavelength, 210 nm.
RESULTS AND DISCUSSION
Separation of the Phthalate Esters. When 0.05 M SDS
solution was used, the phthalate esters migrated in the order
DMP, DEP, DIPP, and DNPP, followed by the other
phthalate esters, all within 20 min (applied voltage, 20 kV;
temperature, 38.9 "C).8 This order is similar to that in GLC,Q
since the separation mechanism in MEKC is based on
distribution between the solvent and the micelle.lP2J0 Theoretical plate numbers of these peaks estimated from uncorrected peak widths were 100 000-180 000 in this MEKC
separation. DMP, DEP, DIPP, and DNPP may be partially
distributed into the micelle, but the other phthalate esters
may be mostly distributed. The distribution must be better
controlled in order to separate more phthalate esters.
Accordingly, we used a mixture of methanol with SDS
solution in order to reduce the distribution of phthalate esters
into the micelle.10 Addition of methanol (20%,v/v) to 0.05
M SDS solution gave better resolution in the separation of
the phthalate esters. The chromatogram is shown in Figure
1. In addition to DNPP and DIPP, the separation of another
pair of alkyl isomers, DNBP and DIBP, was achieved. BBP
and DNAP were also separated. This order is also similar to
that in GLC.9 The theoretical plate numbers of these peaks
estimated from uncorrected peak widths were 170 000-200 OOO
in this MEKC separation, but the migration time became
longer (ca. 60 min). This phenomenon is caused by a decrease
in the electroosmotic velocity.1° The influence of organic
solvents on the electroosmotic velocity has been studied in
the literature;" that is, the electroosmotic velocity and p
potential decrease with increasing content of organic solvent
and this trend is explained by changes in the dielectric
properties of the electric double layer and of the charge
generation on the fused-silica surface. Also, the influence of
methanol on SDS micelle formation has been studied by light
scattering measurements.12 At a mole fraction of methanol
of 0.12 (about 23% methanol (v/v)),the aggregation number
of SDS micelles is lower than that in pure water. But the
(8) Takeda, S.; Wakida, S.; Yamane, M.; Kawahara, A.; Higashi, K.
Anal. Sci. 1991, 7 , 1113 (Suppl).
(9) Ishida, M.; Suyama, K.; Adachi, S. J. Chromatogr. 1984,294,339.
(10) Otsuka, K.; Terabe, S.; Ando, T. Nippon Kagaku Kaishi 1986,
(7), 950.
(11) Schwer, C.; Kenndler, E. Anal. Chem. 1991,63, 1801.
(12) Parfitt, G. D.; Wood, J. A. Kolloid 2.2.Polym. 1969, 229, 55.
=
tR
-
(1)
tO(1 - tR/tmc)
Experimentally, to and ,&
,, were measured with methanol and
the overlapping peak of DEHP and DNOP. Under the
conditions in which the solution containing 20% methanol
was used, it is difficult to choose a marker for the micelle.
Sudan I11 gave more than one peak under these conditions,
and the migration time of the last peak agreed with that of
the peak of DEHP and DNOP. Therefore, as an approximation, the velocity of the micelle is equal to that of DEHP
and DNOP.
As described ref 2, the capacity factor & decreased almost
linearly with an increase in the electroosmotic velocity urn, as
the applied voltage was varied over the range of 5-30 kV. The
cause of the phenomenon is the rise of temperature within
the capillary _dueto joule heating. Consequently, each line
of the plot of k' vs ueowas extrapolated to the intercept a t u,
= 0, and the value of &' a t u, = 0 was assumed to be the &'
a t the temperature of the air flowing around the capillary.
These corrected capacity factors were used in the calculations
that appear below.
The capacity factor &' can be related to the distribution
coefficient K,2
I
KD(c, - cmc)
(2)
where D is partial specific volume of micelle, c ~ is
f the
concentration of surfactant, and cmc is the critical micellar
concentration. The plots of the corrected capacity factor vs
SDS concentration showed good linearity from 0.03 to 0.15
M (Five concentrations were used, and the correlation
coefficientswere over 0.99,). The distribution coefficientwas
calculated from the slope of the plots of &' vs the concentration
of SDS. The value of 0 was calculated from literature data13
at the experimental temperature.
In environmental analysis, the octanol-water partition
coefficient P is an important parameter used to estimate the
behavior of hazardous chemicals in the environment. Therefore, we compared the calculated distribution coefficients in
micellar solubilizationto octanol-water partition coefficients.
The octanol-water partition coefficients in the literature'
were measured a t 20 "C, so the distribution coefficients at 20
"C were estimated from the variation of K with temperature.2
The variation of K with temperature is as follows (the van't
Hoff equation):
k'
l n K = - - AHo ASo
(3)
RT+R
where AHo and ASo are the thermodynamic parameters
enthalpy change and entropy change in micellar solubilization
of the phthalate esters. R is the gas constant and T is the
absolute temperature. The van't Hoff plots of experimental
data showed good linearity over the range of experimental
temperatures (the setup temperatures were 26.7, 38.9,51.1,
and 63.3 OC, and the correlation coefficients were over 0.98).
The distribution coefficients at 20 OC were evaluated by
extrapolating these plots.
(13) Shinoda, K.; Soda, T. J. Phys. Chem. 1963,67,2072.
ANALYTICAL CHEMISTRY, VOL. 65, NO. 18, SEPTEMBER 15, 1993
12
Table I. Calculated and Observed Migration Times of
Phthalate Esters with SDS Solution.
tR(obs)/min
error/ %
solute
tR(calc)/min
DNAPo
DMP
DEP
DIPP
DNPP
9
DIPP,
14.6
19.9
21.2
21.8
14.5
19.3
21.5
21.7
0.7
3.1
-1.4
0.5
Solution, 0.05 M SDS (pH 9.0); applied voltage, 20 kV; tamperature, 26.7 "C. to(obs) = 5.76,t,,(obs) = 22.2.
Y
-C
Table 11. Calculated and Observed Migration Times of
Phthalate Esters with Methanol Mixed Solution.
solute
tR(calc)lmin
tR(obs)/min
error/ %
6
D
3
2491
w
M
v
DMP
DEP
DIPP
DNPP
DIBP
DNBP
BBP
~
I
I 1
I
3
5
log P
Figure 2. Plots of logarithmsof the distributlon coefficients In micellar
solublllzation vs logarithms of octanol-water partklon coefficients of
phthalate esters at 20 "C with SDS solution (a) and methanol mlxed
solution (20%, v/v) (b).
Logarithmicplots of the distribution coefficientsvs octanolwater partition coefficients for both SDS solution and
methanol mixed solution are shown in Figure 2. These plots
for both solutions show good linearity except for DNAP. This
result suggests that the mechanism of separation is almost
the same in both two solutions. With regard to DNAP, it
migrates at almost the same speed as the micelle, so the
calculated value of K for DNAP may include a relatively larger
error than the values of K for other phthalate esters. Also,
according to ref 7, the reported value of the octanol-water
partition coefficient of DNAP is doubtful because of the
impurity of the reagents.
By use of these linear relationships between the distribution
coefficients and octanol-water partition coefficients, the
migration time of phthalate esters may be estimated from
their octanol-water partition coefficients, P. These linear
relationships can be expressed as
In K = a log P + b
(4)
Where a and b are constants that can be determined from the
observed data. Substitution of eqs 1 and 2 into eq 4 gives
t, =
t,
exp(a log P + b + c ) + to
1 + exp(a l o g P + b + c )
(5)
where
+
c = ln@(c,, - cmc)} ln(to/tmc)
The dependence of k' on u, is not considered here. Actually,
the migration orders of other substances sometimes change
with electroosmotic velocity.2
To determine the values of a and b, at least two data sets
of log P and t R are necessary. In the following estimation, we
have provisionally used the values calculated from the plots
shown in Figure 2. The calculated values of t R from the
literature values of log P for the phthalate esters are shown
in Tables I (with SDS solution) and I1 (with methanol mixed
solution). They are nearly equal to the observed values. The
octanol-water partition coefficients were measured at 20 "C,
and the distribution coefficientsat 20 "Cwere estimated from
the dependence of K on the temperature as described above.
However, to, tm,, and observed values of migration time at
26.7 "C were used because the Model 270A cannot control
17.2
30.1
42.0
53.8
70.5
75.3
77.4
16.5
26.5
44.3
49.2
68.7
71.3
73.9
4.2
12.0
-5.5
8.6
2.5
5.3
4.5
Solution, 0.05 M SDS (pH 9.0)-methanol (80:20,vlv); applied
voltage, 20 kV; temperature, 26.7 "C. to(obs) = 9.96,t,(obs) = 80.8.
Table 111. Enthalpy and Entropy Changes in Micellar
Solubilization with SDS Solution
AH"/kJ
ASo/J
AH"/kJ
As"/J
solute
mol-'
mol-' K-l solute
mol-'
mol-' K-'
DMP
DEP
-18.7
-19.7
-21.4
-12.9
DIPP
DNPP
-20.6
-21.8
-3.3
-4.2
Table IV. Enthalpy and Entropy Changes in Micellar
Solubilization with Methanol Mixed Solution (20%, v/v)
AH"/kJ
ASo/J
AHolkJ
Aso/J
solute
mol-l
mol-' K-l solute
mol-'
mol-' K-'
DMP
DEP
DIPP
DNPP
-13.0
-19.4
-21.9
-24.1
-17.1
-28.4
-26.9
-31.3
DIBP
DNBP
BBP
DNAP
-33.4
-37.5
-45.1
-77.7
-49.3
-60.1
-81.9
-168.3
the internal temperature at values below the room temperature + 5 "C.If this calculation is applied to other conditions
or substances, it should be noted that the constants a and b
may be different from those of this case.
Thermodynamic Parameters. The thermodynamic parameters, enthalpy change AH" and entropy change AS", in
micellar solubilizationof the phthalate esters can be estimated
from the dependence of K on the temperature as described
above. The values of enthalpy and entropy changescalculated
from eq 3 are listed in Tables I11 (for SDS solution) and IV
(for methanol mixed solution).
All of the enthalpy changes in Tables I11and IV are negative
and decrease with an increase in the alkyl chain length of
phthalate esters. The value of the n isomer is smaller than
that of corresponding i isomer for the two pairs of isomers.
Therefore, in both solutions, the phthalate esters distributed
in the micelle are more advantageous with regard to enthalpy
than those distributed in the solvent. The phthalate esters
having longer alkyl chains or n isomer are more advantageous
with regard to enthalpy in the distribution of the micelle.
All of the entropy changes are also negative with both
solutions. The phthalate esters distributed in the micelle
are less advantageous with regard to entropy than those
distributed in the solvent. While the entropy changes
increased with an increase in alkyl chain length with SDS
solution, they decreased with an increase in alkyl chain length
with methanol mixed solution. However, the value of the n
2492
ANALYTICAL CHEMISTRY, VOL. 65, NO. 18, SEPTEMBER 15, 1993
isomer is smaller than that of the i isomer with both solutions,
that is, the n isomer is less advantageous with regard to
entropy. This difference in the variation of entropy change
with alkyl chain length for the two solutions can be explained
as follows.
The entropy changes obtained from these experiments are
the values of the total system and can be divided into the
entropy changes of the phthalate esters themselves and that
of the surrounding solvents.14 The entropy changes of
phthalate esters themselves decreased with an increase in
alkyl chain length because the longer the alkyl chains they
have, the more limited their motions are in the micelle. The
entropy changes of surrounding solvents increased with an
increase in alkyl chain length because the longer the alkyl
chains the phthalate esters have, that is, the stronger their
hydrophobicity is, the further the %ystalline" structure made
by the solvent molecules extends when the phthalate esters
are in the solvent.16Jg In the case of SDS solution,the entropy
changes of solvents contribute more than those of the
phthalate esters themselves to the entropy changes of the
total system. On the other hand, in the case of methanol
mixed solution, the entropy contribution of the "crystalline"
structure of the solvent molecules is weakened by the presence
of methanol. Therefore the entropy changes of the phthalate
esters themselves contribute more than those of the solvent
to the entropy changes of the totalsystem. As for the isomers,
it is believed that the entropy changes of solvents are about
the same but the entropy changes of phthalate esters
themselves are different from each other. The reason may
be that both the n isomer and the SDS molecule have straight
alkyl chains.
The thermodynamic data for homologous seriessometimes
suggest that variations in enthalpies are opposed to variations
in entropies in such a way that they compensate for each
other on the free energy. In other words, plots of enthalpy
changes vs entropy changes often form straight lines with a
positiveslope. This is sometimes called the enthalpy-entropy
compensation effect.17J8 This effect has been found in a wide
variety of processes and reaction equilibria, and it has been
suggested that the cause of the phenomenon is solute-solvent
(14) Terabe, S., personal communication.
(15) Tamaki, K. Hyoumen 1966,3,527.
(16) Frank, H.S.; Evans, M.W . J . Chem. PhYS. 1946,13,507.
(17) Krug, R. R.Znd. Eng. Chem. Fundom. 1980,19,50.
(18) Takeyama, N.;Nakashima, K. Nippon Kagaku Kaishi 1987, (4),
610.
r
Y
-
r
E"
-J
2-
\
-1801
-80
-60
-40
AH"/ kJ mol-'
-20
Flguro 3. Plots of enthalpy changes vs entropy changes In mlcellar
solubilkatlon with SDS soluHon (a) and methanol mlxed solution (20 %,
vlv) (b).
interaction in many cases.17 According to the solutesolvent
interaction theory, the enthalpy-entropy compensation effect
is explained as follows. When the enthalpy of a solute is
decreased by solvation, simultaneously the freedom of the
solute is reduced; that is, the entropy of the solute is also
decreased.
Plots of enthalpy changesvs entropy changes obtained from
the analytical data of present experimenta are shown in Figure
3. While the compensation effect is found with methanol
mixed solution, it is not found with SDS solution. The
explanation of the enthalpy-entropy compensation effect
described above does not take intoaccountthe entropy change
of the solvent. Therefore, the experimental data suggest that
the enthalpy-entropy compensationeffect is valid only under
the condition that the parameters of the solvent are negligible.
In the case of SDS solution, the entropy changes of solvents
contribute to a greater extent than when solutions containing
methanol are employed, as mentioned above. Therefore the
enthalpy-entropy compensation effect is not observed. These
considerations indicate that the entropy changes in the case
of SDS solution can be explained by the entropy changes of
the solvents.
RECEIVEDfor review August 11, 1992. Accepted June 5,
1993.